Metraipolis › archive › dc_109abe078073
Predictable Scaling: Efficient Model Exploration via Scaling-Law Prediction — arXiv:2610.23453
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.23453
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-19T09:10:17.410Z
- SHA-256 of the archived text
6b39a5516c2f31a557abe03e0a1e4ecf7ce24a5b2afb0b4b0700f842135a0cdf- Stable URL of this page
https://metraipolis.com/documento/dc_109abe078073
How to cite
Predictable Scaling: Efficient Model Exploration via Scaling-Law Prediction — arXiv:2610.23453 — arXiv (en), https://arxiv.org/abs/2610.23453 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_109abe078073 (Metraipolis archive dc_109abe078073, sha256 6b39a5516c2f31a557abe03e0a1e4ecf7ce24a5b2afb0b4b0700f842135a0cdf)
BibTeX
@misc{dc109abe078073,
title = {Predictable Scaling: Efficient Model Exploration via Scaling-Law Prediction — arXiv:2610.23453},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.23453},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_109abe078073; sha256 6b39a5516c2f31a557abe03e0a1e4ecf7ce24a5b2afb0b4b0700f842135a0cdf},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.23453. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Evaluating candidate changes (hypotheses) to recommendation models requires full-scale training runs consuming thousands of GPU-hours, severely limiting the hypothesis space engineers can feasibly explore. We introduce Predictable Scaling (PS), a systematic framework that exploits empirically verified regularity conditions to predict full-scale model performance from small-scale proxies. PS constructs scaled-down variants via composable scaling functions along distinct resource axes, and projects metrics to the target scale via curve fitting with calibrated uncertainty estimates. We compare three prediction strategies of increasing expressiveness: even the simplest (raw proxy) achieves prediction error within the empirical noise floor of repeated-run variance; the most expressive (Gaussian Process) provides an indicative 15% relative MAE improvement and calibrated uncertainty. Evaluation on two ranking models of increasing complexity---a task-specific model and a foundation model---demonstrates that PS correctly ranks actionable hypotheses across diverse hypothesis types at 79-88% compute reductions. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.23453 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-19T09:06:29.551Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).